On 16 March 2006 at 08:46, Diethelm Wuertz wrote:
| Dear all,
|
| I took some time to look at the code, but I not really know what to do.
| The Fortran
| program concerns the SQP algorithm written not by myself but with extension
| added by me. I had also some problems to run it for the Mac and under
| Windows with the Mingw gfortran/gf77 compilers but I think these were
| (hopefully)
| solved.
|
| I have not more time to spend on the (old) m68k, and I hope we can live
| with it.
| Sorry, and many thanks
Yes, we concluded that it's probably most gcc's fault as Fortran support
really sucks on some architectures (on which no sane person would run
numerically challenging code -- arm, m68k and the like which is why this will
probably never get much better than it is).
Reducing optimisation one level to -O1 fixes the issue as Stephen confirmed.
That's a fair compromise as we can't and wouldn't expect you to rewrite this
is a modern language
Cheers, Dirk
--
Hell, there are no rules here - we're trying to accomplish something.
-- Thomas A. Edison
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
